Association for Public Policy Analysis & Management

In the 2016-17 school year alone, 1.36 million U.S. students were identified as homeless. These students, who are disproportionately Black or Latinx, are at risk of a range of adverse educational experiences and outcomes such as low attendance, poor test scores, and low graduation rates (Cowen, 2017; Masten et al., 2012; Obradovic et al., 2009; Rafferty et al., 2004). Students experiencing homelessness tend to move homes—and schools—frequently. School mobility not only truncates learning but can also disrupt students’ support networks (Miller, 2011; Murphy & Tobin, 2011). The vast majority of these students live below the poverty line—thus they may struggle with food insecurity, inadequate access to medical care, and other poverty-related challenges. Thus, there are inherent challenges in disentangling the impacts of homelessness separate from poverty (Buckner, 2008).

Scholars are also increasingly examining the heterogeneity of student homelessness. Students experiencing homelessness spend their nights in a variety of different places, such as shelters, doubled-up in the dwellings of friends, motels, and unsheltered contexts (NCHE, 2017). While qualitative research highlights the importance of these residential contexts, there is limited quantitative work in this area (e.g., Deck, 2017; Howland, Chen, Chen, & Min, 2017; Low, Hallett, & Mo, 2017). Towards that end, our work seeks to provide rigorous quantitative evidence on the educational outcomes of students experiencing homelessness and explore how their outcomes vary by residential context.

We situate the issue of homeless residential context on student educational outcomes in Houston Independent School District. Houston ISD, the seventh largest U.S. school district, which is over 90% non-white, has nearly 7,000 homeless students annually (a number which quadrupled in the year following Hurricane Harvey). Of these, more than 4 in 5 are living doubled-up: Just 1 in 10 live in shelters, 1 in 20 are unsheltered, and 1 in 30 reside in motels. Using pre-Harvey data, we compare students experiencing homelessness to a matched sample of low-income housed peers with similar student profiles using coarsened exact matching (CEM; Iacus et al., 2011). We attend to a broad range of student outcomes including: attendance, discipline, achievement, and on-time graduation.

Our results highlight the heterogeneity of student experiences of homelessness and their differential impact on student outcomes. For example, while all students experiencing homelessness attend less school than their low-income-housed peers, students who live doubled-up tend to miss fewer days than students in other residential contexts. Overall, students experiencing homelessness had lower achievement than their low-income-housed peers. However, these effects are substantially attenuated after adjusting for differences in attendance.

Interestingly, while students living doubled-up have lower achievement than their low-income-housed peers, students who are unsheltered and residing in motels/hotels tend to outperform their low-income housed peers on achievement. Moreover, we find that students who experienced homelessness in high school were equally likely to graduate on time than their low-income-housed peers. Ongoing analyses seek to understand the ways in which school and student supports as well as exclusion from testing may contribute to findings. We conclude with implications for policy and school leadership.